AIThis post was created with the assistance of artificial intelligence (AI).

By Thorsten Meyer | November 5, 2026 | ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

Three pieces of money news landed within two days this week. On November 3, Australia’s central bank raised interest rates to their highest level since 2008 and wrote, in its official statement, that AI investment is pushing up the prices of technology goods. Later that day in the US, AMD reported that almost two thirds of its revenue now comes from data centers. On November 4, Hong Kong passed a law that makes crypto platforms report their customers’ transactions to the tax office.

These are not one story, and I won’t pretend they are. Two of them are connected, and the connection comes from the central bank itself, not from me. The third is a different kind of signal. Here is what each one says, where the link is real, and where it isn’t.

This article explains public data from company filings, a central bank and a government. It is not investment advice, and it is not tax or legal advice.

Key takeaways

  • Chips: AMD’s data center segment brought in a record $8.5 billion in the third quarter of 2026, up 97% from a year earlier. That is about 64% of AMD’s revenue; a year ago it was 47%. AMD expects data center to be more than 80% of its revenue in 2027. That last number is AMD’s own forecast.
  • Rates: The Reserve Bank of Australia (RBA) raised its cash rate to 4.85% on November 3, the highest level since late 2008, in a split 6 to 3 vote. Its statement says “AI-related investment is driving rapid growth in prices for technology-related goods”.
  • The link: RBA Governor Michele Bullock said AI investment demand is “outstripping savings”, which pushes up the neutral interest rate. In plain words: for now, the AI boom can make money more expensive, not cheaper.
  • Tax: Hong Kong passed the law that brings in the OECD’s Crypto-Asset Reporting Framework. From January 1, 2027, crypto service providers must collect tax data on their customers. The tax office starts sharing it with partner countries in 2028.
  • What doesn’t connect: Hong Kong’s law is about tax transparency, not AI. It is in this piece because it shows the same shift from another side: digital money is becoming as visible to tax offices as a bank account.
Video: three signals on the money behind AI (AMD’s data center quarter, Australia’s rate decision and Hong Kong’s crypto tax law) and what connects them, with animated charts, in about six minutes. Narration: AI voice clone of Thorsten Meyer; images AI-generated.

Products Worth Considering

LG UltraGear 32GS60QC-B 32-inch Curved QHD Gaming Monitor (180Hz, 1ms, 1000R, AMD FreeSync, HDR10)

LG UltraGear 32GS60QC-B 32-inch Curved QHD Gaming Monitor (180Hz, 1ms, 1000R, AMD FreeSync, HDR10)

Bose SoundLink Flex Bluetooth Speaker (2nd Gen) – Portable Outdoor Speaker with Hi‑Fi Audio, Waterproof and Dustproof, USB‑C, Up to 12 Hours

As an affiliate, we earn on qualifying purchases.

Signal 1: Chips. Where the AI money lands

If you want to know where AI budgets end up, look at the companies that sell the hardware. AMD published its third-quarter results on November 3 and filed its quarterly report (10-Q) with the US Securities and Exchange Commission on November 4. The quarter ended September 26, 2026.

  • Total revenue: $13.4 billion, up 45% from a year earlier and up 16% from the second quarter.
  • Data center segment: $8.5 billion, up 97% from $4.3 billion a year earlier and up 27% from $6.7 billion in the second quarter. Operating income in the segment rose to $2.9 billion from $1.1 billion.
  • Share of revenue: about 64%. CFO Jean Hu said it was “up from 58% last quarter”. CEO Lisa Su put it at 47% one year ago.
  • The rest of the business: client and gaming fell 8% to $3.7 billion; gaming alone fell 40%. Embedded chips grew 29% to $1.1 billion.

A note on a number you may have seen. Some headlines before the report said data center “hits 58% of revenue”. That was the second-quarter share. The third quarter came in at about 64%.

AMD’s data center segment covers AI accelerators and server processors, plus network and programmable chips. AMD names strong demand for its EPYC server processors and the launch of its Instinct MI450 accelerators as the drivers. It says Anthropic, Meta, Microsoft, OpenAI and Oracle Cloud Infrastructure are running models and workloads on its first Helios rack-scale systems.

The forecasts are bigger still. AMD expects fourth-quarter revenue of about $17 billion, plus or minus 3%. In its prepared remarks, it said it expects server revenue to double in 2027, total revenue to grow by more than 80%, and data center to make up more than 80% of revenue. These are company forecasts, not results.

Close-up of a round silicon wafer with a shimmering grid of tiny chips resting in a white holder on a bright lab bench
Where AI budgets land: about 64% of AMD’s revenue now comes from its data center segment. AI-generated illustration.

Three details from the filings show the other side of the boom:

  • Building ahead of demand. Inventory rose by about $3 billion in a single quarter, to $11.5 billion. Free cash flow was $554 million. In its 10-Q, AMD warns that it keeps entering long-term purchase commitments “in advance of demand”, which raises the risk if its demand estimates turn out wrong.
  • Customers who can become shareholders. AMD has issued warrants to OpenAI and Meta for up to a combined 320 million AMD shares, subject to vesting and exercise conditions. Chipmaker and AI labs are tied together more tightly than a normal supplier relationship.
  • Even the winner pays for money. In August, AMD borrowed $4.75 billion through new bonds with interest rates between 4.6% and 5.5%. A company growing this fast still pays a real price for capital.

And the boom has a cost for ordinary buyers. AMD said PC sales to end customers slowed late in the quarter “as higher memory and component costs weighed on demand”. It is planning for PC market shipments to fall by a double-digit percentage in 2027. Why memory got so expensive is a story of its own; we covered it in The AI Boom Has Reached Your Shopping Basket.

Thorsten Meyer AI
Signal 1 · Chips

AMD · third quarter 2026 · quarter ended 26 Sep

Data centers now bring in almost two thirds of AMD’s revenue

AI budgets turn into chip revenue at record speed. The forecasts are bigger still, and so are the bets AMD makes ahead of demand.

64%
The number to remember
Share of AMD’s revenue that came from its data center segment in the third quarter of 2026, up from 47% a year earlier.
$8.5BData center revenue, Q3 2026 (record)
+97%Data center, year over year
$13.4BAMD total revenue, Q3 2026
~$17BAMD outlook for Q4 2026 (forecast, ±3%)

Data center share of AMD revenue

Per quarter; 2027 is AMD’s own forecast

  • Q3 2025a year earlier47%
  • Q2 2026the “58%” in pre-earnings headlines58%
  • Q3 2026reported 3 Nov 2026~64%
  • 2027forecast, not a result>80%

The other side of the boom

From the 10-Q and the prepared remarks

  • Built ahead of demandInventory up about $3B in one quarter, to $11.5B
  • Customers as shareholdersWarrants to OpenAI and Meta for up to 320M shares combined, with conditions
  • Money has a price$4.75B in new bonds, August 2026, at 4.6% to 5.5%
  • PCs pay tooAMD plans for PC shipments to fall by a double-digit percentage in 2027; memory costs weighed on demand

Sources: AMD earnings release, 3 Nov 2026 · AMD prepared remarks, 3 Nov 2026 · AMD 10-Q, 4 Nov 2026. Data center segment = AI accelerators, server CPUs, network and programmable chips. Not investment advice.
Powered by Thorsten Meyer AI

Signal 2: Rates. What the AI boom does to the price of money

On November 3, the RBA’s Monetary Policy Board raised the cash rate target by 0.25 percentage points to 4.85%, effective November 4. It is the fifth increase this year: the rate started 2026 at 3.60%. According to the RBA’s own table of rate changes, the target was last higher in late 2008, when it stood at 5.25% until a cut on December 3, 2008. That is where the “18-year high” headline comes from.

The vote was split. Six members voted for the increase. Three wanted to hold at 4.60%.

AI is not the main reason for the increase. The Board said inflation is still high and widespread, energy prices remain high because of the conflict in the Middle East, and it sees “an increasing risk that higher inflation expectations will become embedded”. It now expects inflation to come back to the top of its target band in late 2027, “around three quarters later than previously forecast”.

But AI shows up in the statement three times:

  • “AI-related investment is driving rapid growth in prices for technology-related goods”.
  • Business investment is expected to be stronger in the near term, “including data centre-related investment”.
  • Growth among Australia’s trading partners beat expectations because “the boost from AI-related investment has outweighed the adverse effects of the Middle East conflict”.

At the press conference, a reporter asked when AI would start to help bring inflation down. Bullock’s answer is the most important sentence of the week for anyone who borrows money. She said AI investment demand is so large that “it’s outstripping savings, and that’s leading to an increase in what we call the neutral interest rate.” And on the supply side: “We’re only getting the impact on demand.”

The neutral interest rate is the level that neither speeds the economy up nor slows it down. If it rises, a given rate brakes the economy less than before, and central banks may need higher rates for longer to get the same effect. The productivity gains that would let the economy grow faster without inflation may come later. They haven’t arrived yet.

Bullock also added two caveats of her own. First, “we think maybe about 70 per cent is imported” when it comes to data center investment, and she said estimates vary. That imported share doesn’t strain Australia’s own economy. Second, she made no promise about the next step: “I don’t know what the next move will be.”

Data centers are financed with a mix of cash, bonds and long leases, so a higher price of money feeds straight into what they cost to build. We looked at that financing in Who Pays for the AI Data Center Boom?.

Thorsten Meyer AI
Signal 2 · Rates

Reserve Bank of Australia · 3 November 2026

Australia’s rate hits an 18-year high, and the central bank names AI

The neutral rate is the level that neither speeds the economy up nor slows it down. If AI investment pushes it up, money stays expensive for longer.

4.85%RBA cash rate target from 4 Nov 2026
6–3Board vote; three wanted to hold at 4.60%
5Increases in 2026 (start of year: 3.60%)
2008Last time the rate was higher

The cash rate in 2026

Effective dates, per the RBA’s table

  • 1 Janstart of 20263.60%
  • 4 Feb+0.253.85%
  • 18 Mar+0.254.10%
  • 6 May+0.254.35%
  • 30 Sep+0.254.60%
  • 4 Nov+0.25 · highest since 20084.85%

Bar length relative to 5.25%, the level of November 2008.

Where AI shows up

In the statement and at the press conference, verbatim

  • Prices“AI-related investment is driving rapid growth in prices for technology-related goods”
  • InvestmentStronger in the near term, “including data centre-related investment”
  • Imports“we think maybe about 70 per cent is imported”, Governor Bullock on data center investment
  • Supply“We’re only getting the impact on demand”, Bullock

“It’s outstripping savings, and that’s leading to an increase in what we call the neutral interest rate.”RBA Governor Michele Bullock on AI investment demand, media conference, 3 Nov 2026

AI is one factor, not the main reason: the Board names broad inflation, high energy prices and the risk that inflation expectations become embedded.

Sources: RBA statement, 3 Nov 2026 · RBA media conference, 3 Nov 2026 · RBA cash rate table. Not investment advice.
Powered by Thorsten Meyer AI

Signal 3: Tax. Who gets to see digital money

On November 4, Hong Kong’s Legislative Council passed the Inland Revenue (Amendment) (Crypto-Asset Reporting Framework and Amended Common Reporting Standard) Bill 2026. The government welcomed the vote the same day. The law brings two standards from the OECD, the club of mostly rich countries that writes international tax rules, into Hong Kong law:

  • CARF, the Crypto-Asset Reporting Framework (published by the OECD in 2023). Businesses that exchange crypto for their customers collect tax data once a year. The tax office then exchanges that data automatically with partner countries.
  • The amended Common Reporting Standard (CRS). Hong Kong has exchanged bank-account data under the CRS since 2018. The update brings in digital money products such as central bank digital currencies and certain e-money products, derivatives that reference crypto, and investment funds that invest in crypto.

The timeline, from the government and the Inland Revenue Department:

  • January 1, 2027: the CARF rules apply. Service providers start collecting the data.
  • 2028: first exchanges with partner jurisdictions. The amended CRS takes effect on January 1, 2028.
  • 2029: first exchanges under the amended CRS.

What gets reported: three types of transactions, namely crypto against regular money, crypto against other crypto, and transfers of crypto, including transfers to outside wallets. The report covers the customer’s name, address, countries of tax residence, tax ID number, date and place of birth, and yearly totals per type of crypto-asset.

Who has to report: businesses that exchange crypto for customers and have a link to Hong Kong. The tax office lists exchanges, brokers, dealers and crypto ATM operators as examples. They must register on a new reporting portal even if they have nothing to report. During the bill’s review, the government agreed that a provider with no reportable transactions only files a nil return. A customer who knowingly or recklessly gives a provider a false self-declaration on their tax residence faces a fine of up to HK$10,000, according to the Legislative Council committee report.

Christopher Hui, Hong Kong’s Secretary for Financial Services and the Treasury, said implementing the latest standards “is crucial to maintaining Hong Kong’s reputation as an international financial and commercial centre.”

An open blank ledger book with a magnifying glass on a pale blue desk, a harbour city skyline blurred in the window behind
From 2027, crypto platforms linked to Hong Kong collect tax data on their customers. AI-generated illustration.
Thorsten Meyer AI
Signal 3 · Tax

Hong Kong · crypto tax reporting

Hong Kong opens crypto accounts to tax offices

The third signal is not about AI. It shows governments catching up with where money now moves. Below: what links the three signals, and what doesn’t.

  1. 22 May 2026Bill publishedFirst reading in the Legislative Council on 3 June
  2. 4 Nov 2026Law passedImplements the OECD’s CARF and the amended CRS
  3. 1 Jan 2027CARF appliesCrypto service providers start collecting customer tax data
  4. 2028First exchangesTax office shares CARF data with partner jurisdictions; amended CRS takes effect
  5. 2029Amended CRS exchangesNow including digital money products and crypto funds

What gets reported

Per customer, once a year

  • Crypto against regular money
  • Crypto against other crypto
  • Transfers, including to outside wallets
  • Name, address, tax residence, tax ID, date and place of birth, yearly totals per crypto type

Who reports

Businesses that exchange crypto for customers, with a link to Hong Kong

  • Exchanges, brokers, dealers, crypto ATM operators
  • Register on a reporting portal, even with nothing to report
  • No reportable transactions: a nil return
  • False self-declaration by a customer: fine up to HK$10,000

1Chips

AI budgets become chip revenue: 64% of AMD’s quarter.

Link to AI: direct

2Rates

AI investment lifts demand and tech prices before productivity.

Link to AI: direct, in the RBA’s words; one factor of several

3Tax

Digital money becomes visible to tax offices across borders.

Link to AI: none directly

Sources: HK Government, 4 Nov 2026 · Inland Revenue Department, CARF · LegCo Bills Committee report. Not tax or legal advice.
Powered by Thorsten Meyer AI

What connects the three, and what doesn’t

SignalWhat it showsLink to AI
Chips (AMD)AI budgets turn into chip revenue at record speedDirect: accelerators and server chips
Rates (RBA)AI investment lifts demand and tech prices before it lifts productivityDirect, in the RBA’s own words; one factor among several
Tax (Hong Kong)Digital money becomes visible to tax offices across bordersNone directly; it is a transparency signal

The first two form a chain. Money flows into AI hardware, and AMD’s numbers show how fast. That demand shows up in prices and in how a central bank thinks about rates. Then money costs more for everyone else, including the companies building data centers and the people with a mortgage.

What I am not claiming: AMD’s results did not cause Australia’s rate decision. The RBA set its rate for Australia’s economy, mostly for domestic reasons. And Australia’s cash rate does not set what AMD pays on its US dollar bonds.

The third signal runs in parallel. It is not about AI. It is about governments catching up with where money now moves. Put together, the week says something simple: the AI boom is no longer only a technology story. It now shows up in central bank statements, in what a laptop costs, and in the rules for who can see your money.

What this means for companies

  • Treat compute as a scarce input with a rising price. AMD says customers expanded their deployment plans and it is securing supply far in advance. If your AI plans depend on hardware, lock in capacity early and budget for price increases, including for ordinary PCs.
  • Plan with today’s cost of money. If AI investment pushes neutral rates up, cheap capital is not around the corner. AI rollouts and data center projects that rely on borrowed money need a business case that works at current rates.
  • Watch concentration. Customer warrants and multi-year supply deals tie chipmakers and AI labs together. Spreading AI workloads across more than one supplier reduces that dependency for you.
  • Ask for the productivity, not just the spending. The RBA’s point is that AI has lifted demand but not yet supply. Boards should ask the same question of their own AI budgets: where is the measured gain?
  • Crypto on the balance sheet or in payments. If your company or your customers use crypto platforms with a link to Hong Kong, expect requests for tax-residence self-declarations from 2027. Get that data in order early, and ask your tax adviser what applies to you.

What this means for users

  • Gadgets: The AI boom competes with you for memory chips. AMD expects PC shipments to fall by a double-digit percentage in 2027, and it says higher memory and component costs already weighed on demand.
  • Borrowers: If you live in Australia and have a variable-rate loan, the November increase reaches your repayments. The bigger point applies everywhere: if the neutral rate rises, lower rates may take longer to return.
  • Crypto users: If you use a platform covered by Hong Kong’s rules, expect a self-declaration form asking for your tax residence and tax ID. From 2028, that data can flow to the tax office of your home country, if it is a partner jurisdiction. Giving false information knowingly or recklessly can bring a fine.
  • Everyone: AI is not only something you use. Through prices and interest rates, it is now something you help pay for.

Honest limits

  • AMD’s figures for the fourth quarter and 2027 are company forecasts. Results can differ, and AMD lists its risk factors in the 10-Q.
  • AMD’s data center segment is not only AI. It also includes server processors, network chips and programmable chips. AMD does not split out a pure AI share in the segment figures.
  • AI is one factor in the RBA decision, not the main one, and Australia’s situation is specific: in Bullock’s words, demand there already exceeds supply. Bullock’s 70% import estimate is the RBA’s, with uncertainty she named herself.
  • “Highest since 2008” is based on the RBA’s table of cash rate changes; CNBC reported the same.
  • Hong Kong has not yet published its lists of reportable and partner jurisdictions, and the tax office’s detailed guidance is still to come.
  • We used no share prices and no crypto prices. Nothing here is investment, tax or legal advice.

Sources

Related on Thorsten Meyer AI: The Market Just Sent Big Tech Its AI Bill looks at how investors started to price the cost of the AI build-out.

You May Also Like

Zoom AI Companion 3.0: What Its November Launch and $12 Custom Add‑On Mean for Business

AIThis post was created with the assistance of artificial intelligence (AI).Executive summary.Zoom’s…

Your AI Answer in a Stranger’s Session: Cloudflare’s KV Cache Leak and the AI Spam Drowning Bug Hunters

Cloudflare’s hosted GLM-5.3 model returned 155 responses containing other customers’ data after a memory-pool misconfiguration met an SGLang bug. Meanwhile, AI-written bug reports flood security teams: Firedancer now charges 1 SOL per report. What happened, what a KV cache is, what the reports leave open, and what companies and users should do.

Chinese Video AI Models Beat US Rivals In Visual Quality, But Lag On Safeguards: HKU Study – South China Morning Post

A headline about an HKU study says Chinese video AI models beat U.S. rivals on visual quality but lag on safeguards; study details remain unavailable.

How to Use Claude Mythos for Complex Reasoning Tasks

Discover how Claude Mythos from Anthropic pushes AI reasoning to new heights. Learn how to use it effectively and what makes it stand out.